Generative Multimedia

What it is about

You will learn the fundamentals and modern methods of generative modeling with a focus on multimedia (images, text, music). In the end you will understand the principles, architectures and applications of VAEs, GANs, diffusion models, Transformers and energy-based models, and you will be able to implement and train simple models as well as critically assess current research.

What you will be able to do

• Understand principles and architectures of modern generative models
• Know the strengths, limitations and applications of VAEs, GANs, Transformers, diffusion and energy-based models
• Implement generative models for images, text and music and design training strategies
• Critically evaluate state-of-the-art research on multimodal models and large language models

What the module consists of

• VorlesungIntroduction to theory, concepts and applications; interactive with discussions and Q&A
• Labor / PraktikumHands-on implementations and programming-related exercises, in sync with lecture topics

Teaching method

• Präsentationen und VorlesungsfolienConveying theoretical concepts and context
• Interaktive Vorlesungssitzungen (Q&A, Diskussionen)Deepening and clarification of content
• Interaktive Coding-SessionsApplying what was learned in practical implementations
• Diskussionen von ForschungspapierenCritical engagement with current research
• Gastvorträge von Branchenexpert:innen (falls verfügbar)Practical relevance and industry insights
• Tutorien mit ProgrammieraufgabenIndependent practice and implementation
• Essays zu ForschungsarbeitenSummarizing and critically evaluating literature